Skip to Content

How to Create a Blog Post Outline Using ChatGPT (and MarketMuse)

6 min read

Guess what?

There’s a better way to do this! Read Using MarketMuse AI to Create Content and watch the accompanying video.

One line prompts aren’t going to get you very far. In this post I’ll show you a way to construct a better outline prompt for ChatGPT, but you can apply the process when creating any prompt.

Now, if you’re using the MarketMuse ChatGPT integration, you’re already ahead of the game because MarketMuse data gets inserted into the prompt. But I’m going to show you how to turn it up notch.

Sure, the internet is full of resources featuring simple one-sentence prompts. But try them yourself and you’ll find the output to be very uninspiring. That’s just the way large language models (LLMs) like ChatGPT work.

They are always determining the most likely next word, given what’s come before. So if you use a simple prompt, chances are you’ll end up with some basic output. There will be slight variations from time to time, since LLMs are probabilistic and not deterministic, but the results won’t be anything earth shattering.

Whether you use generative AI to create an article or not, you still need an outline. An LLM still needs direction just as much as a human writer does. Because as I’ve already said, the output from one-line prompts are very pedestrian.

So let’s get started.

Prompt Engineering for Content Marketers

Here’s the four-step process I use. It’s a framework that you can employ to create prompts for any content marketing task:

  1. What I find works well in creating a prompt is first asking ChatGPT what input it needs to create the desired output. For example, I literally asked “What information do you need to create an outline for a blog post?” and it gave me a short list of half a dozen items.
  2. Then I ask what else it needs. I repeat this process until the answers start going off the rails or get repetitive.
  3. I clean that list of responses so that I have something concise to work with.
  4. Then I use that cleaned list to interrogate ChatGPT as to whether or not including those items will have a material effect on the output. At the same time I’m looking to get a sense of how it’s been trained on these concepts in the list. Essentially, I’m looking to understand its limitations.

Although the process appears simple enough, it can require significant time and effort. In one instance, I went down the rabbit hole of determining if ChatGPT can analyze the content of a given URL. Long story short it can’t (at least at the time of writing) — but it took me a while to confirm it. ChatGPT excels at convincingly making stuff up!

OpenAI also has a great resource, best practices for prompt engineering and I encourage you to take a look at that. Although it’s geared more towards their API, there are some technical details that apply to ChatGPT as well.

Prompt Engineering Guide is another valuable resource. It goes more in depth and has a broader application than the OpenAI article. Here’s a good cheat sheet too, for writing more effective prompts.

A Prompt for Creating Blog Post Outlines With ChatGPT

Let’s skip ahead to the good stuff and look at a ChatGPT prompt for creating blog post outlines. It’s a variation on the MarketMuse Create Outline prompt that I created using the same process mentioned above. Here’s what it looks like.

ChatGPT prompt for an article outline along with numbered references to the major components describe further in the blog post.
ChatGPT Prompt for an outline.

Here are the components that make up the prompt and why you should use them:

  1. Topic: identify the main topic of your blog post. Refer to this post on keywords vs topics if you’re having trouble differentiating between the two.
  2. Purpose: inform/persuade/entertain/inspire — these are one’s I’ve confirmed that ChatGPT can distinguish between.
  3. Target audience: also include their knowledge level, and why they’re searching using this term as this can have a material impact on the structure.
  4. Position in the marketing funnel: post-purchase, bottom-of-funnel-middle-of-funnel, or top of funnel are concepts that ChatGPT is acquainted with.
  5. Word Count: length of content can influence an outline, particularly in terms of the number of subheadings — shorter pieces of content naturally tend to have fewer.
  6. Tone: conversational/humorous/educational/inspiring — ChatGPT can discriminate between these different tones of voice.
  7. Style: professional/storytelling — like tone of voice, I include these to ensure they are reflected in the subheadings and not just the final draft.
  8. Format: listicle/how-to/case study/thought leadership/product review — ChatGPT can distinguish these differences in format and I include this to ensure the proper output.
  9. Angle: decide on the angle or perspective that you want to take in your blog post, such as a contrarian viewpoint or a unique insight. This is a great opportunity to differentiate your content!
  10. Outcomes: define the outcomes or takeaways that you want readers to gain from the article. Defining the purposes helps steer the output.
  11. Related topics: MarketMuse list of semantically related topics to influence the outline.

So, what does this look like in the end? Here it is!

Comparison of a Detailed Prompt vs a One-Sentence ChatGPT Prompt

Let’s compare the output of two different prompts. The output on the left is from the detailed prompt as shown above while the right-hand side shows the results of a single-sentence prompt.

Comparison of the output of a detailed ChatGPT prompt and a simple one-sentence prompt.
Output comparison of a detailed ChatGPT prompt vs. a simple one.

I’ll let you decide which you prefer, but here are a few observations.

Output from the one-sentence prompt appears to be much more generic. It’s no surprise given that the only information provided was a topic and request for an outline. That’s what you get when you provide no guidance. Whether you make the request of a human or a machine is irrelevant.

Just look at the subheadings choices — planting tomatoes, tomato plant care, pests and diseases. These all sound like run-of-the-mill topics.

The story is a lot different when it comes to the detailed prompt. All those added elements (purpose, position, tone, style, etc.) substantially affect the output. Keep in mind that you’ll want to carry those elements over to the content itself, when it comes to writing. Whether it’s ChatGPT or a human that’s responsible, both will appreciate the added direction.


Simple one-sentence ChatGPT prompts are available everywhere you look. But using these generic prompts brings generic results — it’s a race to the bottom.

With the proliferation of ChatGPT prompts, it’s easy to feel like there’s no need to be creative or use any nuance in your interactions. After all, simple one-sentence prompts are everywhere. But using these generic prompts brings generic results — it’s a race to the bottom.

You can avoid this situation when using generative AI and it all starts with how you craft your prompt.

Stephen leads the content strategy blog for MarketMuse, an AI-powered Content Intelligence and Strategy Platform. You can connect with him on social or his personal blog.